Atmospheric Motion Vector observations in the ECMWF System: Fourth year report

TitleAtmospheric Motion Vector observations in the ECMWF System: Fourth year report
Publication TypeReport
Date Published05/2015
Series/CollectionEUMETSAT/ECMWF Fellowship Programme Research Reports
Document Number36
AuthorsSalonen, K, Bormann, N
Event Series/CollectionEUMETSAT/ECMWF Fellowship Programme
Place of publicationShinfield Park, Reading

EUMETSAT has introduced several changes to the processing of AVHRR AMVs from Metop-A and Metop-B satellites resulting to improved data quality. EUMETSAT has also made available a new dual Metop-A/B AMV product. It is the first AMV product with global coverage. The latest monitoring results are discussed in the report. Investigations with AMVs giving coverage over the Indian Ocean have been carried out. Currently Meteosat-7 is the prime satellite over that area. Other satellites giving coverage over the region are the CMA operated FY-2D and FY-2E, and IMD operated Kalpana-1 and INSAT-3D. The first monitoring results for INSAT-3D are generally in line with what is seen for other GEO satellites, suggesting promising data quality. The quality of FY-2E AMVs has improved during recent years, and it also shows promising data quality. Results from an impact study with Meteosat-7 and FY-2E AMVs indicate neutral to positive impact for both satellites. The situation dependent observation errors are used in the ECMWF operational system from cycle 40r1 onwards. In the context of reanalysis activities the estimates for height and tracking errors are not always available. The impact of using default values for the height and tracking errors has been investigated. The results indicate mainly neutral impact compared to using the more tailored height and tracking error estimates. The work on alternative interpretations of AMVs continues. A set of data assimilation experiments has been performed to investigate the impact of using a layer averaging observation operator. Also the impact of re-assigning the AMV height based on model best-fit pressure statistics has been considered. Using the traditional single-level observation operator together with the height re-assignment indicates positive forecast impact. Experimentation with layer averaging gives more mixed results.

PDF icon Download (4.01 MB)